olly - Fotolia


Use and understand a Hyper-V failover cluster with these tips

To build single or multisite Hyper-V failover clusters, you need to meet the proper requirements, choose appropriate hardware, and understand the inner workings of the cluster.

Hyper-V failover clusters are complex systems, designed to protect VMs in the event of a failure. If the clusters are not built correctly with the proper framework, you risk losing all of that data and wasting money and resources. Look no further if you want to learn how to build a Hyper-V failover cluster for one site or even multiple sites. There are also tips for sizing hardware for your failover cluster. Then, take a step inside the cluster to see what's actually going on in there. If after all that, you're still unsure whether Hyper-V failover clusters are the right approach for your company, keep reading for some effective alternatives. These five quick links are just what you need to get started.

Setting up a Hyper-V failover cluster

Despite popular belief, it is possible to make your VMs highly available using the free Hyper-V Server. In order to do this, though, you'll not only need a copy of Hyper-V Server; you'll also need a storage array, an understanding of how to deploy and configure a failover cluster and familiarity with PowerShell. Once you've installed the Hyper-V Server and established the initial configuration, you just need to use a few simple commands and connect your shared storage to the cluster, and you should be good to go. One thing to keep in mind, however, is that if you decide to build a free Hyper-V failover cluster, you will still incur a cost associated with licensing VMs.

Build a multisite failover cluster

It's important to add more hosts, in addition to having powerful servers, to reduce your overall waste percentage.

There are a number of hardware and storage requirements involved in building a multisite Hyper-V failover cluster to protect VMs against a network outage, power outage or even a full-blown natural disaster. In addition, when it comes to data replication for cluster storage, you'll have to decide between synchronous replication, which has the potential for reduced system performance, and asynchronous replication, which has the potential for data loss. Another thing to take into consideration is how you will configure the quorum's voting process, which determines whether a cluster remains online. Finally, be prepared to reconfigure the VM's IP address when failover from one site to another does occur.

Size hardware for your failover cluster

Although buying the most powerful servers you can is an obvious approach to effectively supporting Hyper-V failover clusters, RAM oversubscription limits complicate the purchasing process. Not only do you need to reserve some RAM for host processing, but you must also have enough residual and unused RAM in the remaining hosts for the lost host's VMs. The key is to minimize wasted RAM. It's important to add more hosts, in addition to having powerful servers, to reduce your overall waste percentage. Start by calculating how many VMs you have and how many gigabytes of RAM will be allocated to each VM to figure out how many hosts you'll need.

Hyper-V failover cluster
Failover when both nodes are running Hyper-V.

The failover cluster in action

In order to achieve high availability, each VM needs to have its own resource dynamic link library (DLL), called VMCLUSRES.DLL, which interacts with VMs in the Hyper-V failover cluster via Microsoft Cluster Service (MSCS), and more specifically, Resource Host System (RHS). The VMIsAlive function, executed every five seconds by VMCLUSRES.DLL, detects when the VM needs to be restarted on another node. Acting as the messenger between the VM and MSCS, RHS receives the VM's status from VMCLUSRES.DLL and delivers it to MSCS. When a failover is necessary, MSCS has Failover Cluster Manager take it from there. Through these constant status updates, resource DLL protects all of the VMs in the Hyper-V failover cluster.

Alternatives to a Hyper-V failover cluster

If you can't afford to build a failover cluster, there are other options. The good news is that Microsoft doesn't require a Cluster Shared Volume in Windows Server 2012, so even without using shared storage, you can still provide fault tolerance. You just need enough local storage to handle the VMs. You could also use Microsoft Hyper-V Replica, which takes the virtual hard drives that make up VMs and replicates them from one host to another. The Hyper-V Replica feature allows you to create replicas only using a source server and a target server, which is more cost-effective than the three cluster nodes or two cluster nodes and a file share witness needed for a failover cluster.

Next Steps

Establish secure failover and restores

Failover cluster pitfalls addressed

Take advantage of Hyper-V Replica

Dig Deeper on Virtualized clusters and high-performance computing